Contaminated raw cookie dough wasn’t on anyone’s mind as my public health colleagues and I were searching for the cause of a multistate outbreak of E. coli infections.

I’m one of the Epidemic Intelligence Service officers in CDC’s Enteric Diseases Epidemiology Branch, which monitors and investigates foodborne diseases together with CDC’s Enteric Disease Laboratory Branch and state health departments. On any given day we are working on several clusters and outbreaks of illness.

In mid-May, CDC’s PulseNet Team alerted us about a cluster of E. coli O157 infections. We began working with state and local health departments to investigate these infections. We originally suspected ground beef, which is one of the “usual suspects” for E. coli O157, along with leafy greens and sprouts. As the labs in states and at CDC found more and more people infected with the same strain, the demographics shifted; patients were generally young and female, which isn’t what is normally seen with ground beef-associated outbreaks.

We got copies of the interviews on standard questionnaires that state investigators did with ill people and looked through them for other suspicious food sources, but nothing was conclusive. None of the food items implicated in past E. coli O157 outbreaks appeared to be associated with this one. Therefore, we decided to conduct what we call “open-ended hypothesis-generating interviews,” in which we call the people affected and just talk about everything that they had eaten and done the week before they became ill, looking for things in common among them. Standard questionnaires are useful, but they are only asking for answers to a series of questions. Sometimes something with a broader scope, like this sort of wide-ranging interview, is needed to find things that are unusual and might not be asked on our questionnaires.

Washington State was kind enough to let CDC do the interviews on their five patients. Mark Sotir and I reached the mother of the first patient on a Saturday. She mentioned that her child had eaten raw prepackaged cookie dough during the days before he got sick. On Sunday, I reached a second patient, and she told me she had eaten at an ice cream shop and had ice cream with cookie dough and brownie mix-ins.

Cookie dough? When cases three, four, and five all confirmed that they ate raw cookie dough, it appeared we had a surprising new possible culprit in our outbreak. (It wasn’t until later that we learned that the second patient also had eaten raw cookie dough at home.)

During an outbreak investigation, we hold a series of multistate conference calls in which CDC and affected states share what we’re finding. Representatives from many of the affected states were on our June 16, 2009 conference call, and I mentioned my cookie dough hypothesis. On the face of it, cookie dough was the most unlikely culprit, but epidemiologists in several other states said, “Oh, yes, I had a case mention that, too”. It became a “Eureka” moment for the group.

At the end of the call we agreed that cookie dough, strawberries, fruit roll-ups, apples, and ground beef were all possible causes. Time to go back to the cases and ask more questions!

A lot of our work is like that. Our branch chief, Patricia Griffin, sometimes says there is a certain “head banging quality” to what we do. It can take many, many interviews and requires a wide-ranging curiosity to consider all the possibilities.

There are no short cuts. We talk to the patients, we look at the combined information, and we generate hypotheses about the cause. Then we can refine our questions and go back to the patients again to see which hypothesis holds true.

After the conference call we created a short-form questionnaire and states quickly interviewed their case-patients. Within a day we had enough confirmatory responses that a case-control study was conducted. The state investigators interviewed people who were of similar age and gender as the cases but who weren’t sick with E. coli during the same time frame about what they had eaten, and we compared them with the people who were sick. We call these people “matched controls”. With each step, the association between cookie dough and illness became stronger and stronger. By the end of the week, there was a recall on Nestle Toll House refrigerated cookie dough.

If you asked me how I arrived at cookie dough as a possible cause, it was because we were really looking for anything that could be the reason people were getting sick.

Talking with the patients, sifting through their answers, and being willing to consider any option, including those where the connection just seems almost impossible, and close collaboration with epidemiologists and laboratorians in the states, turned out to be the keys to finding the answer.